Definition of RAID
-
Hi all,
i'm a little confused over RAID.
- Data protection of Hard Drive.
Do a complete backup on a regular basis. RAID (Redundant Array of Inexpensive Disks) - two or more drives used in combination.
- Disk mirroring
- Disk striping.
*edit*
the definition i got for disk striping from my lecture was "dividing files into pieces and putting each piece in other drives"
not very clear to me.
*/edit*
Can someone explain what does RAID do as well as disk mirroring and disk striping?
Thanks
-
Originally posted by 93'guy:
Hi all,
i'm a little confused over RAID.
- Data protection of Hard Drive.
Do a complete backup on a regular basis. RAID (Redundant Array of Inexpensive Disks) - two or more drives used in combination.
- Disk mirroring
- Disk striping.
*edit*
the definition i got for disk striping from my lecture was "dividing files into pieces and putting each piece in other drives"
not very clear to me.
*/edit*
Can someone explain what does RAID do as well as disk mirroring and disk striping?
Thanks
RAID is a collection of various methods to use multiple drives like they were one single unit, whether to stripe, mirror, or a combination.
Disk mirroring, or RAID 1, allows for protection over a disk failure. It takes 2 drives, and data is written exactly the same way in both drives. Basically, Drive 1 and Drive 2 share the same contents, so should 1 drive fail, you have an exact backup in the other drive.
Disk Striping, or RAID 0 (More like AID 0, there's nothing redundant about this), takes 2 drives, and combines them as if they were one big unit. It writes different sections of data into the different drives. Take a data file for instance. It is broken up into 6 parts, A,B,C,D,E and F, by the RAID controller. Then you have 2 drives in (R)AID 0, D1 and D2. The controller writes A,C and E to D1 and B,D and F to D2. So, when you are reading the file again, the controller can read A and B simultaneously, the C and D, the E and F.(R)AID 0 increases the capacity and speed of the array, but also the risk. Should D1 fail in the array, all data is lost because without parts A, C or E stored in D1, the file cannot be read because it is incomplete.
There is also RAID0+1, and RAID10 (or 1+0). They both require 4 drives. It basically takes 1 RAID 0 array and 1 RAID 1 array. With 0+1, 2 drives are first striped. Once the RAID 0 array has been created, the controller uses the other 2 drives to mirror the original RAID 0 array. 1+0 is the opposite, first mirroring, then striping. The differences, I'm not sure, but apparently they are different enough to make controller manufacturers list them both out.
RAID 5 is a hybrid of 0 and 1. However, it requires 4 drives instead of 2. Lets say you have 3 data files, A, B and C, each split into 3 parts. A1, B1, C1 and D(P) are written to D1. A2, B2, C(P) and D1 are written to D2. A3, B(P), C2 and D2 are written to D3. A(P), B3, C3 and D3 are written to D4 X(P) is a parity bit, allowing the system to rebuild data parts should 1 drive fail. How it does this, I don't know. The parity bit can also be constructed using 3 data parts, should the drive with the parity bit go down. The advantages over RAID 1 is that you get increased performance over RAID 1, as well as increased capacity. RAID 1 loses half the total capacity of the array, but RAID 5 loses the capacity of 1 disk in the array. Lets say you built a RAID 1 array with two 2TB drives. You effective data capacity is 2TB. With RAID 5, however, using four 1TB drives, you have an effective capacity of 3TB.
-
RAID - Redundant Array of Independent Disks.
Basically at basic user level careful selection of the RAID type to be used assure that in the event of one of your disk breaking down, another disk will still function. Mostly employed in corporate business storage and servers systems. But since several years ago such basic functions are already present in motherboards for consumers. Thus driving makings of PCI RAID cards out of business.
-
Originally posted by troublemaker2005:
RAID - Redundant Array of Independent Disks.
Basically at basic user level careful selection of the RAID type to be used assure that in the event of one of your disk breaking down, another disk will still function. Mostly employed in corporate business storage and servers systems. But since several years ago such basic functions are already present in motherboards for consumers. Thus driving makings of PCI RAID cards out of business.
Actually, there still is demand for PCI/PCI-E based RAID controllers, such as those by Adaptec/LSI/Intel RAID controllers, meant for enterprise solutions. These cards offer much better throughput, security and reliability than onboard controllers.
Also, some chipsets, like Intel's H55 and other boards, do not support RAID. Therefore, those who wish to use RAID can purchase an inexpensive SATA RAID controller.
-
Originally posted by Raraken:
RAID is a collection of various methods to use multiple drives like they were one single unit, whether to stripe, mirror, or a combination.
Disk mirroring, or RAID 1, allows for protection over a disk failure. It takes 2 drives, and data is written exactly the same way in both drives. Basically, Drive 1 and Drive 2 share the same contents, so should 1 drive fail, you have an exact backup in the other drive.
Disk Striping, or RAID 0 (More like AID 0, there's nothing redundant about this), takes 2 drives, and combines them as if they were one big unit. It writes different sections of data into the different drives. Take a data file for instance. It is broken up into 6 parts, A,B,C,D,E and F, by the RAID controller. Then you have 2 drives in (R)AID 0, D1 and D2. The controller writes A,C and E to D1 and B,D and F to D2. So, when you are reading the file again, the controller can read A and B simultaneously, the C and D, the E and F.(R)AID 0 increases the capacity and speed of the array, but also the risk. Should D1 fail in the array, all data is lost because without parts A, C or E stored in D1, the file cannot be read because it is incomplete.
There is also RAID0+1, and RAID10 (or 1+0). They both require 4 drives. It basically takes 1 RAID 0 array and 1 RAID 1 array. With 0+1, 2 drives are first striped. Once the RAID 0 array has been created, the controller uses the other 2 drives to mirror the original RAID 0 array. 1+0 is the opposite, first mirroring, then striping. The differences, I'm not sure, but apparently they are different enough to make controller manufacturers list them both out.
RAID 5 is a hybrid of 0 and 1. However, it requires 4 drives instead of 2. Lets say you have 3 data files, A, B and C, each split into 3 parts. A1, B1, C1 and D(P) are written to D1. A2, B2, C(P) and D1 are written to D2. A3, B(P), C2 and D2 are written to D3. A(P), B3, C3 and D3 are written to D4 X(P) is a parity bit, allowing the system to rebuild data parts should 1 drive fail. How it does this, I don't know. The parity bit can also be constructed using 3 data parts, should the drive with the parity bit go down. The advantages over RAID 1 is that you get increased performance over RAID 1, as well as increased capacity. RAID 1 loses half the total capacity of the array, but RAID 5 loses the capacity of 1 disk in the array. Lets say you built a RAID 1 array with two 2TB drives. You effective data capacity is 2TB. With RAID 5, however, using four 1TB drives, you have an effective capacity of 3TB.
Thanks for the explanation!
I kinda get it now
-
Originally posted by Raraken:
Actually, there still is demand for PCI/PCI-E based RAID controllers, such as those by Adaptec/LSI/Intel RAID controllers, meant for enterprise solutions. These cards offer much better throughput, security and reliability than onboard controllers.
Also, some chipsets, like Intel's H55 and other boards, do not support RAID. Therefore, those who wish to use RAID can purchase an inexpensive SATA RAID controller.
I agree. But those were for smaller desktop or smaller server systems. Corporate or companies running on large storage capacities usually get 1U ~3U or more form factor external storage systems as backups for servers. These external storage systems have their own backplane with SCSI, Fibre channels and SATA interfaces and in RAID arrays.